On Thu, Jan 07, 2016 at 10:31:27AM +1100, Dave Chinner wrote:
> From: Dave Chinner <dchinner@xxxxxxxxxx>
>
> For large sparse or fragmented files, checking every single entry in
> the bmapbt on every operation is prohibitively expensive. Especially
> as such checks rarely discover problems during normal operations on
> high extent coutn files. Our regression tests don't tend to exercise
count
> files with hundreds of thousands to millions of extents, so mostly
> this isn't noticed.
>
> However, trying to run things like xfs_mdrestore of large filesystem
> dumps on a debug kernel quickly becomes impossible as the CPU is
> completely burnt up repeatedly walking the sparse file bmapbt that
> is generated for every allocation that is made.
>
> Hence, if the file has more than 10,000 extents, just don't bother
> with walking the tree to check it exhaustively. The btree code has
> checks that ensure that the newly inserted/removed/modified record
> is correctly ordered, so the entrie tree walk in thses cases has
these
> limited additional value.
>
> Signed-off-by: Dave Chinner <dchinner@xxxxxxxxxx>
> ---
> fs/xfs/libxfs/xfs_bmap.c | 10 ++++++++--
> 1 file changed, 8 insertions(+), 2 deletions(-)
>
> diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
> index 7388495..bc7e7d5 100644
> --- a/fs/xfs/libxfs/xfs_bmap.c
> +++ b/fs/xfs/libxfs/xfs_bmap.c
> @@ -325,9 +325,11 @@ xfs_check_block(
>
> /*
> * Check that the extents for the inode ip are in the right order in all
> - * btree leaves.
> + * btree leaves. THis becomes prohibitively expensive for large extent count
This
Reviewed-by: Brian Foster <bfoster@xxxxxxxxxx>
> + * files, so don't bother with inodes that have more than 10,000 extents in
> + * them. The btree record ordering checks will still be done, so for such
> large
> + * bmapbt constructs that is going to catch most corruptions.
> */
> -
> STATIC void
> xfs_bmap_check_leaf_extents(
> xfs_btree_cur_t *cur, /* btree cursor or null */
> @@ -352,6 +354,10 @@ xfs_bmap_check_leaf_extents(
> return;
> }
>
> + /* skip large extent count inodes */
> + if (ip->i_d.di_nextents > 10000)
> + return;
> +
> bno = NULLFSBLOCK;
> mp = ip->i_mount;
> ifp = XFS_IFORK_PTR(ip, whichfork);
> --
> 2.5.0
>
> _______________________________________________
> xfs mailing list
> xfs@xxxxxxxxxxx
> http://oss.sgi.com/mailman/listinfo/xfs
|